Congratulations to the authors for the compilation of the CAMELS-DK dataset and the high-quality documentation and provision of extensive information in this manuscript.

CAMELS-DK is a very valuable addition to the family of CAMELS datasets, with the incorporation of groundwater data being a milestone for large-sample hydrological datasets.

I have attached my comments as a PDF file, although there are many comments, most are minor or technical or suggestions for improvement, the quality of the manuscript and dataset is already very high.

The manuscript describes the CAMELS-DK dataset, which consists of hydrometeorological time series and catchment attributes for a large sample of catchments in Denmark. I appreciate the valuable contribution of the authors in extending the CAMELS dataset family to Denmark and providing a valuable dataset for the hydrological community. However, the manuscript requires substantial revision before it can be published.

One of the major concerns is the lack of observed streamflow data for many of the catchments. As described in the manuscript, the observed streamflow data is available only for the 304 catchments out of 3330 mentioned in the Title. A key characteristic of CAMELS datasets is providing observed streamflow data – without this, the datasets are less useful for large-sample hydrological analyses. I appreciate that the authors have included a modeled flow time series. However, I recommend keeping 304 catchments in the Title, i.e., where the authors have calculated hydrological signatures based on the observed data.

Specific comments:

1) Title: The authors should keep the number of gauged stations (i.e., 304) in the Title of the manuscript to be consistent with other CAMELS datasets.

2) Abstract: Line 17-22 can be rephrased to clearly distinguish that CAMELS-DK provides dynamic and static variables, including observed streamflow and their temporal range for 304 catchments, and provides dynamic and static variables for an additional 3026 catchments for further applications. Also, can the author mention how the additional catchments can enhance applicability without observed streamflow data? I understand that modeled streamflow is provided, but I am not convinced about its applicability given many catchments are very small in area, limited model resolution, and missing validation.

3) Line 44-48: Please include the access dates of the cited weblinks.

4) There are several typos throughout the manuscript. I am restricting myself from providing a complete list here. However, the authors should thoroughly proofread the manuscript to correct the typos and grammatical mistakes.

5) Line 103: “consistent data for 3300 catchments”? – As described in the manuscript, all data is not consistent for all 3330 catchments. Maybe “consistent meteorological and static attribute data for 3330 catchments? Moreover, is it 3300 or 3330? Please be specific.

6) Line 126: Please include a reference for ID15v2.5. Also, if ID15v2.5 includes 3351 catchments, why do the authors include 3330 in the manuscript? Given that many of these catchments are very small, how do authors justify the reliability of the model performance with 500 m resolution data [Line 208]?

7) There are several mistakes in the figures and visualization. For example, Figure 1(b): “Catchment area” instead of “Area”; Figure 2(a): the size of the circles is enlarged in the legend; Figure 2(c) is not colorblind-friendly; Figure 3(a): “ele_mean” instead of “DEM”.

8) Line 359-360: “CAMELS-DK provides observed streamflow data from 304 hydrological stations and simulated streamflow data for 2,942 catchments” – which makes it a total of 3246 catchments?; and it mentions, “However, 388 catchments lack observed or simulated streamflow data.” – then why are those relevant/included in the CAMELS-DK dataset?

9) Can authors include more discussion on the DK-Model performance, especially the causes of deteriorated performance in the central and northern regions? Since the authors provide a modeled streamflow based on this model, the readers/users must understand the limitations.

10) Line 437-438 and Line 450: The Python script for processing the time series, landscape attributes, and original raster data is not included in the dataset. I suggest including the Python script for data processing or adding more details on data processing, detailing the different products and steps used in generating the dataset to build user confidence. I would also recommend adding the license/disclaimer as a text file within the dataset to ensure this is readily available when a user directly downloads the product.

11) There are several mistakes in the naming convention in the dataset. For example, CAMELS_DK_soil.csv uses column name “Id15_model” instead of “catch_id”; “CAMELS_DK_signature_obs_based.csv” and “CAMELS_DK_signature_sim_based.csv” includes the same data. I suggest checking all the files and making necessary corrections thoroughly.